India

Of all the former colonies in Asia, India might have seemed the least likely to develop into a patronage democracy. The dominant Congress party was highly professionalized, led by upper-middle-class lawyers and doctors. Its national leaders seemed genuinely motivated by the needs of their country, all of them following the moral example of M. K. Gandhi in spending years in British gaols during the anti-colonial struggle. The national leadership was committed to modernization and professionalism, seeking inspiration in the diverse developmental histories of the Soviet Union, the United States, Japan, and of course Britain. India's new leaders also inherited the remnants of the famed Indian Civil Service (ICS). The ICS was the pioneer of the Weberian bureaucracy: professionalized, well-paid, and free from political intermediation at the level of implementation. Indeed, the model for civil service education in Britain came from the colonial civil service college at Haileybury. Why then did patronage become so deeply entrenched after independence?

The institutionalization of a localized form of patronage was a result of the imperatives of state building in the context of strong centrifugal and disintegrative pressures. The Congress hierarchy faced a set of contradictory political imperatives in the 1940s; first, ironically, was the need for national unity in the face of growing centrifugal threats to the integrity of the state wrought by decolonization;Footnote ⁶⁴ second was the need to secure a base of resources to support the Congress party's provincialized electoral machine that had grown strongly since the 1930s. The persistence of patronage over time also derives from this fragmented state structure. Indeed, the further de facto decentralization that followed Nehru's death in 1964 increased the power of the state bosses and along with it the entrenchment of corruption in Indian politics.Footnote ⁶⁵ All efforts to reform the system thus far have floundered.

The provincialization of Indian government was centuries in the making. While I begin my account with the introduction of provincial government in the interwar period, one could go back not just to the piecemeal conquest of the subcontinent by the British, but to the prior conquest by the Mughals, and to the fragmented system of warring kingdoms that preceded it in turn.Footnote ⁶⁶ The reason I begin the story in the twentieth century is that this was a critical period of institutional development in Indian history and the one in which provincialization became formalized. At the turn of the century, administrative control over British India was firmly in the hands of the colonial governor. Provincial offices were directly subordinate to the capital, but by the early twentieth century, reluctant to cede administrative control, the British instead granted limited representative government at the provincial rather than central level. The introduction of limited franchise elections at the municipal, district, and tehsil (county) levels under the so-called Morley–Minto reforms (1909) consolidated the power of an emerging class of professional politicians. Until the First World War, politics remained an elite affair but the war effort and the economic boom and bust that accompanied it greatly politicized the Indian countryside. Under the Government of India Act (1919), the British Government introduced a form of government known as diarchy, under which it would retain central control over subjects like defense, law and order, land revenue, and irrigation, while elected ministers in the provinces would have control over local government, health, education, and agriculture. Despite the limited nature of these reforms, they had important consequences. The control of the capital over the provinces was loosened and the latter gained access to land revenues. Brown writes that ‘the provincial arena of politics was greatly enriched by new resources, and its ground rules were changed by new institutions.’Footnote ⁶⁷ Provincialization was further strengthened with the passage of the Government of India Act (1935). With its accession to provincial office in 1937 in seven out of eleven provinces in British India, Congress began to focus on building a mass party.Footnote ⁶⁸ However, the construction of multiple provincial units rather than a cohesive national one would have important implications when it came to the state-building process after independence.Footnote ⁶⁹ The elected leaders of states were loosely tied, if at all, to the leadership of the Congress.

It is crucial to note that Nehru, Patel, and the rest of the Congress Working Committee (CWC) were not just gripped by a fit of absent-mindedness when it came to the issue of bureaucratic autonomy.Footnote ⁷⁰ To keep the country together, the Congress leadership had to balance the need for central control with the imperatives of electoral politics. Although from our contemporary perspective, the Balkanization of India might seem implausible, in the mid-1940s this was a real prospect. Not only did Muslim League demands for autonomy within a federal India threaten to completely emasculate the center, but the prospect of leading princely states like Hyderabad, Travancore, and Jammu and Kashmir opting for independence was far from remote.Footnote ⁷¹ Although the provinces of British India hardly saw independence as a first choice, these centrifugal pressures gave them significant leverage. The violent partitions of Punjab and Bengal only heightened the anxiety of India's national leadership. The center was simply too dependent on the backing of powerful provincial bosses to push a harder bargain. With neither the capacity nor the inclination to pursue universalistic policies, the Congress leadership instead built its party by integrating multiple layers of local patrons who would in turn mobilize their clients.Footnote ⁷² To retain its dominance of national planning and administration and stave off more radical demands for regional autonomy, the center had to concede control over policy implementation to the states.

It was against the backdrop of center–state posturing and provincial politicking that the Constituent Assembly of India was formed.Footnote ⁷³ India is often described as quasi-Federal or as having a ‘unique’ federal structure. Certainly, there appears to be much greater central control than in the United States or other federations. Yet, here there is still much misunderstanding. The Constitution retained many of the key elements of the provincial structure put in place in 1935, in particular, the continued devolution of control over most areas of social policy to the states. With the passing of the Constitution, subjects of government were divided between the center and the states. The center retained control over key policy issues.Footnote ⁷⁴ With respect to loans and grants for development-related expenditure, states frequently complained that they were not allowed enough latitude.Footnote ⁷⁵ Yet, in practice, implementation was strictly a state affair. This gave the Chief Ministers of India's states enormous leverage over the direction of social policy after independence. There was thus significant latent potential for ‘regional bossism’ even if Nehru's charismatic authority kept it in check until his death in 1964.Footnote ⁷⁶

But if Shefter is correct, why was the politicization of the bureaucracy not prevented by the formal institutional rules governing its autonomy? By the late interwar period, India had a conspicuously professionalized bureaucracy that retained its independence vis-à-vis the Indian political elite. Public service commissions (PSCs) governing appointments to the ICS were introduced first at the All-India level in 1923, and then in the provinces throughout the 1930s.Footnote ⁷⁷ Given these institutional safeguards of bureaucratic autonomy, and Home Secretary Vallabhbhai Patel's impassioned defense of the civil service, it has often been assumed that the ICS, the so-called ‘steel frame’ of the British Empire in India, survived decolonization more or less intact. However, decolonization had a profound effect on the administration. To begin with, one of the infrequently mentioned facets of decolonization was the physical decimation of the ICS. As Potter has convincingly argued, the administration in some of the more peripheral areas of British India was already being rolled up by in the 1930s.Footnote ⁷⁸ After the onset of the Second World War, this already strained situation was put under further pressure. Over the course of 1946–48, fully 60 percent of the ICS departed. As an example of this hollowing out, consider the United Provinces. Of the 178 ICS members in place at the end of 1945, just eighty-four remained after Independence, of whom twenty-seven were on deputation to the Government of India, service in another state (one), or on extended leave (all British servants pending retirement).Footnote ⁷⁹

Yet, while the need to rebuild the capacity of the civil service rapidly might have compromised its quality, there is no necessary reason why this should have led to its political capture. For this we have to look to the relative strength of the provincial and national political units. On 20–21 October 1946, Sardar Patel convened a Conference of Provincial Premiers. G. B. Pant, the premier of Uttar Pradesh,Footnote ⁸⁰ India's largest state and with its massive representation in the Lok Sabha the effective kingmaker of Indian states, objected to central control over the provincial civil service. He wanted the center to have control only over recruitment and training, and demanded ‘full control of the provinces over the service’.Footnote ⁸¹ Pant and Prakasam wanted full disciplinary authority to vest in the provinces; they noted, ‘All India services are acceptable to us only when control over them gets vested in the state government. The responsibility of the Central government should be restricted to recruitment and post-entry training…’Footnote ⁸² The premiers of Bengal and Punjab also pushed hard for the provincialization of the civil service, on the grounds that ‘This would ensure proper representation of communities and greater attention to local problems and on the whole be conducive to an efficient administration.’Footnote ⁸³ While the provincial elites conceded to Patel the continued existence of the all-India service, they retained extensive discretionary powers. First, the all-India services were limited to just two of the nine originally proposed. Thus, most spheres of state activity, especially in the area of public works, would remain firmly in the hands of the provincial elite. Second, although recruitment to the IAS was largely taken out of the political sphere, ‘Decisions taken on many other matters were demonstrably respectful of the provincial susceptibilities and left substantial autonomy with the provinces.’Footnote ⁸⁴ Among other issues, this applied to cadre strength, the rules of selection for the provincial service quota, benefits, leave and terms of service, and allotment of officers. Despite making considerable concessions as regards the formal structure of the administrative services, the political leadership of the states would thus attain their objectives in other ways. While IAS members could not be arbitrarily dismissed, they could be transferred to another posting for any reason, and they could be suspended on one third pay for extended periods.Footnote ⁸⁵ For the chief minister and cabinet of a state in the Indian Union, these were significant sticks to hold over civil servants. In this way disagreeable civil servants were easily marginalized and the rest soon made to toe the line.Footnote ⁸⁶ Thus, while there were formal rules that sought to preserve the integrity and neutrality of the ICS, in practice, this often did not occur.

In order to keep Congress's multiple factions within the party fold, the national leadership conceded the principle of provincial control over the key elements of the patronage machine: the bureaucracy and government expenditure. What occurred was the parallel development of an increasingly bureaucratized center but a progressively more politicized periphery. The deep penetration of patronage into Indian politics was the result of the imperatives of postcolonial state building in the face of strong centrifugal and disintegrative pressures, which in turn led central leaders to concede control over implementation to provincial and local politicians. The result has been the stubborn persistence of patronage politics despite India's significant economic growth, improvements in health and literacy, increased lower caste political participation, administrative decentralization, and robust inter-party political competition in recent decades.Footnote ⁸⁷

Crucially, reform of this system has proven extraordinarily difficult. Even where there appears to have been political will in the capital to reform the system from time to time, the center has been virtually powerless to control corruption in the states. Throughout the 1950s and 1960s, commissions of inquiry were launched in the attempt to improve the quality of governance. Yet, with little interest in really taking on the party bosses, Nehru largely turned a blind eye to corruption at the state level. Indira Gandhi attempted to remake the political system in the 1970s, not least through the imposition of authoritarian rule from 1975 to 1977. Yet, in the end, her attempt at centralization in the face of highly entrenched local power centers in the states was a failure. The introduction of quotas, or reservations, across the public sector in the 1990s opened up new sources of patronage and corruption that could be exploited by state leaders.Footnote ⁸⁸ The new round of decentralization at the sub-state level in the same period similarly created many new brokers, albeit at a smaller scale.Footnote ⁸⁹ It is true that since 2005 there exists Right to Information (RTI) legislation that could in theory limit the scope for politicians and bureaucrats to engage in corrupt activity. However, it remains unclear whether the autonomy of the enforcement agencies will persist and whether they have the capability to tackle higher level corruption. More generally, while new federal programs utilizing information technologies may lower corruption in some arenas,Footnote ⁹⁰ patronage remains generally rampant in the public sector.Footnote ⁹¹

Ceylon

If the entrenchment of patronage democracy in India appears as the norm in former colonies, the relatively muted development of patronage in Ceylon in the 1950s and 1960s is a notable exception. There is little doubt that political interference in the bureaucracy increased over the course of the 1960s. As in India, even though recruitment was largely a matter for the impartial civil service commission, promotions and transfers within the administration became subject to political influence. Politicians also used their influence to distribute ‘pork’ to their home constituencies. Moreover, from the early 1970s, formal changes in the rules governing bureaucratic autonomy precipitated a marked decline in the quality of governance in Ceylon (by then renamed Sri Lanka). Yet, as Johnston has made clear, corruption is relative. As he notes, even Germany, that paragon of Weberian bureaucratic professionalism, experiences substantial corruption albeit of a different sort to that of, say, Mexico.Footnote ⁹² The point to stress here is that even though Ceylon experienced the slow growth of patronage in the administration over the early decades of independence, this development was muted by comparison to India, Pakistan, Malaysia, the Philippines, or most of its southern Asian neighbors.Footnote ⁹³

When comparing Ceylon with the likes of India, Brazil, or Russia, the significantly different scale of the country has to be borne in mind. Ceylon had an estimated population of just 889,584 in 1827, and even though this had grown to over 6.5 million by the time of independence in 1948, this pales in comparison to India's 319 million (1941). However, these different dimensions are in some ways an important part of the story. Ceylon's small size made centralized government both possible as well as desirable. With much lower demographic pressure and a relatively successful agro-export industry, Ceylon did not experience the same degree of peasant unrest as India in the 1920s and 1930s. An ambitious policy of peasant resettlement enacted in the 1930s was designed to create a sizeable class of peasant owner-cultivators who would form the backbone of the conservative political movement. Ceylon's relatively healthy finances and its crucial role as a British naval base in the Second World War also allowed for the development of some of the best public health and education facilities in colonial Asia by independence.Footnote ⁹⁴ Ceylon certainly had many advantages then. However, Ceylon remains an intriguing case. The relative absence of patronage in postcolonial Ceylon was not predetermined by the nation's small stature. Like India, Ceylon was a hierarchical society with massive inequality between landowners and the landless. Why traditional patrons quickly lost their role after independence remains a puzzle.Footnote ⁹⁵ Ceylon also poses a problem for other theories of fragmented or divided government. It also emerged from independence with significant internal ethnic and class tensions. From the mid-nineteenth century, Indian Tamil laborers had been brought to the island in their thousands to work on the plantations of northern Ceylon. While ‘native’ Tamils, who had migrated to the island centuries previously, were accorded full citizenship, the Tamil laborers were regarded as temporary residents, much as the Chinese were in Malaysia. Thus, the potential for an ethnically divided federal compromise (or even partition) existed. Yet, Ceylon retained its unitary and centralized government. To explain this, we have to look to the much greater centralization of colonial Ceylon and the relative lack of centrifugal and disintegrative pressures at the time of independence.

Ceylon was Britain's ‘model colony.’ The British gradually brought the whole island under direct control by 1815, including the hold-out Kandyan kingdom.Footnote ⁹⁶ The island was originally subordinated to the East India Company administration on mainland India but eventually became a crown colony in its own right under the administration of a Governor-General. The Ceylon National Congress (CNC), modeled on the Indian National Congress, was not formed until 1919. Even more than the Indian Congress, it was the vehicle of the conservative elite. The first representative institutions were introduced in 1921. However, the CNC had no figure comparable to Gandhi who sought to tie the party so directly to the fate of the masses. The more radical caucus within the CNC called the Young Lanka League had only 124 members as late as 1921.Footnote ⁹⁷ By 1924, the franchise had been expanded to 189,335 (about 4 percent of the population) and following the so-called Donoghmore Constitution of 1931, the franchise was extended to all males over 21 and all females over 30.Footnote ⁹⁸ As a result, by the 1930s, ‘The stage was set for the transformation from elite constitutionalism to mass politics.’Footnote ⁹⁹

The institutional structure resulting from the Donoghmore Constitution (1931) was unusual. A unicameral legislature was grouped into seven committees, governing different policy areas. External policy was excluded and reserved for the British Governor. The result was competition among elites for control of the national government, with political competition taking the form of genuine debates over the substance of social policy rather than over the mere distribution of spoils. Critically, the greater centralization of the colonial government concentrated political elites on the capital rather than on their constituencies. Conservative CNC leaders D. B. Jayatilaka and D. S. Senanayake led the colonial government of 1931. From 1931 to 1939, Jayatilaka was the de jure and de facto leader, although after 1940 his power and influence were waning.Footnote ¹⁰⁰ Although the CNC and its successor the United National Party (UNP) were electorally dominant, genuine competition existed in the form of the socialist Lanka Sama Samaja Party (LSSP), which had mobilized considerable political support by the eve of the Second World War.Footnote ¹⁰¹ As an indicator of the competitiveness of the system, within a decade of independence, power had transferred to the opposition.

Interestingly, James Manor cites the lack of local autonomous government in postcolonial Ceylon as a major weakness of the system. He writes that ‘the leadership at the national level has maintained a high degree of centralisation in the political system…’Footnote ¹⁰² Emulating the Westminster parliamentary system, the national political elite were concentrated on the capital, Columbo. Manor's concern was that this made politicians inattentive to the varied needs of the population, favoring top-down rather than bottom-up solutions. Even as independence approached, Ceylon's capital accounted for a proportion of the population more like a developed state than a colony. Ceylon's early leaders worked hard to prevent any decentralization of authority. Owing in part to their aristocratic backgrounds, CNC leaders were reluctant to over-politicize the masses. Although they remained skeptical about the devolution of political power, the CNC, like the Indian Congress, knew by the 1930s that their own ‘representativeness’ depended on the improvement of the lot of the native peasantry. However, they saw their relationship with the peasantry in corporatist or businesslike terms rather than patron–client ones. While there were many large-scale landowners among the political leadership, those in the ruling CNC were predominantly involved in the plantation sector, and thus saw themselves as having employees rather than tenants. Their preference was thus for top-down legislative solutions rather than the decentralization of patronage. The opposition held more or less the same attitude. S. W. R. D. Bandaranaike, Ceylon's fourth prime minister (1956–59), viewed local officials and politicians with suspicion and radically curtailed their autonomy. The central government entrusted bureaucrats rather than local representatives to deliver goods and services to citizens.Footnote ¹⁰³ Government departments were hierarchical and focused on the ministers in the capital. This feature did lead Ceylonese politicians to be somewhat unresponsive to their constituents (especially Tamils), but it had the benefit of forcing them to compete over policy rather than spoils. Indeed, unusually among former colonies, Ceylon developed robust interparty competition based around a fairly recognizable left–right distinction from the mid-1950s. The efficiency and quality of Ceylon's public services was a model in the postcolonial world.

Although Manor argues that centralization and the disenfranchisement of the Tamil population should be understood as two sides of the same coin, I am not convinced. In this respect, Manor is adhering to the view that decentralization, or federalism, is the preferable institutional solution in multi-ethnic polities.Footnote ¹⁰⁴ In fact, the jury remains out on this question.Footnote ¹⁰⁵ In some cases, formal decentralization has exacerbated separatist tendencies and been instrumental in the ratcheting up of violent domestic conflict.Footnote ¹⁰⁶ More important for the hardening of Tamil–Sinhalese ethnic divisions than centralization per se was the deliberate disenfranchisement of Tamil laborers by the conservative Sinhalese political elite. The laws on citizenship that were written into the Constitution disenfranchised migrant Indian Tamils, giving the plantation owners relatively free rein to favor Sinhalese tenants in their social and economic policies. State land was allocated to Sinhalese tenants and landless laborers but not to their Tamil counterparts who were laboring on plantations.Footnote ¹⁰⁷ While this was a vote-winning policy for the Sinhalese elite, by politicizing the distribution of state resources along ethnic lines, these ethnic divisions were themselves politicized in the process.Footnote ¹⁰⁸ Had centralized policy been more neutral in practice, a less conflictual path through the postcolonial period might have been possible.

Ceylon began independence with a uniquely favorable institutional setting. It had a relatively autonomous bureaucracy that largely survived independence. Unlike India, or any of the other states in Southern Asia, Ceylon had a fully indigenous civil service by the time of independence.Footnote ¹⁰⁹ Moreover, unlike the Philippines, Ceylon avoided Japanese occupation during the war with the result that this bureaucracy survived largely intact. Thus, although India inherited the same laws governing bureaucratic autonomy, the opportunity for political intervention following the postwar decimation of the bureaucracy and its rapid post-independence buildup probably made politicization easier in the Indian case. Nevertheless, in Ceylon, the bureaucracy grew much more slowly after independence and it remained relatively free from political intervention at higher levels.Footnote ¹¹⁰

However, the Ceylonese case also illustrates how path dependence does not mean historical determinism. With the institutional reforms of the 1970s the quality of governance in Ceylon began its precipitous decline. Ironically, Ceylon's relatively centralized polity made the decline all the more rapid. Following the suppression of the Janatha Vimukthi Peramuna (JVP) uprising in 1971, the government used emergency rule to centralize authority within the Executive. The PSC, which had governed recruitment to the bureaucracy, was abolished. The constitutional reforms of 1972 and those of 1978 facilitated the politicization of the administration.Footnote ¹¹¹ The latter established a presidential system and extended terms to six years. The lack of institutional checks and balances in the new system, compared with that it replaced, brought about a decline in the quality of governance. Sri Lanka's contemporary quality of governance is thus lower than might be predicted by the theory. However, the (albeit bloody) termination of the Tamil insurgency has again made reform possible, even if the short-term prospects are unclear.Footnote ¹¹² There is nothing in the theory that implies that centralized states will necessarily attempt reform. However, the theory would be strongly supported if subsequent reforms proved relatively effective in Sri Lanka.

Data and Method

Below I conduct some relatively simple ordinary least squares (OLS) regressions on data from the former British Empire. The dependent variable is the persistence of patronage, while the main independent variable is the extent of centrifugal pressures at the time of decolonization. I control for a number of potential historical confounders, but this kind of observational study cannot establish causality in a conclusive way.Footnote ¹¹³ However, the simple conditional correlations presented below reveal some interesting and somewhat counterintuitive insights. First, states that had the most decentralized governing structures during colonial rule and faced the most intense centrifugal pressures at the time of decolonization have had the most persistently high levels of patronage. Second, and perhaps more surprisingly, this relationship is not dependent on the prior level of development of these colonies; that is, many of the more developed colonies have struggled most to decrease corruption and establish effective bureaucracies.

As I noted above, state building is an ongoing process. Here I look specifically to the colonial world, where the period of institution and constitution formation can be isolated with relative precision. In this section, I attempt to capture the disintegrative pressures faced by state builders during the period of decolonization. Even this is not straightforward. One option might be to see whether colonies fought with the imperial power for their independence. Such a contentious birth might be associated with disintegrative pressures. Yet, anti-colonial struggles were often instrumental in bringing colonies together as nations. Indeed, it could be argued that colonies that were effectively handed independence without a nationalist struggle were the least cohesive.Footnote ¹¹⁴ Thus, here I focus on the structure of the late colonial state itself. From an analytical point of view, a nice feature of colonial rule is that it varied significantly in form. While the Iberian colonizers were notorious for ruling through caciques or oligarchs, the French tended to rule much more directly. In fact, at one point parts of North Africa were nearly incorporated into France proper. Yet, comparisons between the colonies of different empires are problematic because of the great deal of underlying and unaccounted for variation across the empires themselves. Thus, I take advantage of institutional variation within the late British Empire.

The British Empire utilized both direct and indirect rule in its colonies. In the former case, a largely British administration would rule a territory, providing security, managing the economy, and covering some social services. In the latter case, the British ruled through native elites. They would either chose or create local leaders who would be tasked with ruling according to the restrictions of British suzerainty. Where colonies were ruled directly in their totality, states were more cohesive. The political battle that followed independence would be over control of the colonial state and its administrative apparatus. In contrast, where colonies were ruled indirectly, multiple elites could either claim to be the legitimate rulers of the new state in its entirety, or they could demand independence for their own territories. In short, indirect rule should be associated with greater disintegrative pressures at the time of decolonization. We have some reasonably good data on the extent of indirect rule within the British Empire in the period immediately after the Second World War. Matthew Lange has created an index of direct rule by measuring the proportion of court cases that are handled by colonial rather than indigenous courts and by the size of the colonial police force.Footnote ¹¹⁵ I use this measure of the intensity of Direct Rule to proxy for the extent of centrifugal pressures at the time of independence. The greater these pressures, the more likely the localized institutionalization of patronage as a way to maintain state cohesion. It needs to be stressed that the mechanism here is the effect of this colonial state structure on decolonization, rather than its direct transmission to postcolonial development. Of course, direct rule may be capturing several other possible mechanisms. Below I try to rule out the alternative claim that indirect rule simply resulted in places that were already more likely to be corrupt. Ultimately, while results from the tests below cannot confirm the mechanism put forward here, we can at least determine whether they rule it out.

Patronage is extremely difficult to measure directly and especially to do so in a way that lends itself to cross-national and intertemporal comparison.Footnote ¹¹⁶ Perhaps the closest measure of patronage comes from Evans and Rauch, who used surveys in order to determine the extent of political interference in the normal operation of the bureaucracy. Unfortunately, the spatial coverage of their data is limited mostly to developed states.Footnote ¹¹⁷ Bo Rothstein and his colleagues have carried out expert surveys in a sample of fifty-two countries (to date) on the impartiality of government and the quality of administration. Of particular relevance is the measure of bureaucratic professionalism (or impartiality). While in the future we hope to be able to use this dataset to provide a more robust evaluation of the effect of decolonization on post-independence governance, for now the problem is that the data is limited in geographic coverage.Footnote ¹¹⁸

Although direct measures of patronage with sufficient geographical coverage are hard to come by, a number of other indices of the quality of governance exist that might be used to proxy for the extent of patronage. Indices tracking the perception of corruption are problematic, especially when attempting to explain temporal change, but they are strongly correlated with more direct measures of patronage. Transparency International's Corruption Perceptions Index (CPI) and the World Bank's Worldwide Governance Indicators are useful here. Rothstein's measure of bureaucratic professionalism is highly correlated with the CPI (r=0.712, p<0.01) and World Bank index (r=0.710, p<0.01), while Evans and Rauch's somewhat older indices of bureaucratic professionalism are both also well correlated with these broader corruption indices. Thus, they provide a useful, if blunt, tool to get at the question. Here I use the World Bank's Control of Corruption Estimate as the primary dependent variable because of its broader geographical and temporal coverage.Footnote ¹¹⁹ I have changed the sign of the index so that a higher level of corruption is associated with a higher number. Another problem here is that even this index is very limited in its temporal coverage. Data for the indices begin in 1996 for most states. Thus, we cannot use them to conclusively demonstrate the continuity of levels of patronage over time. Rather, here we have to make the assumption that most former colonies began with fairly high levels of corruption and that contemporary variation comes in large part from variation in the success of subsequent reforms. In any case, indices of the perception of corruption are problematic for intertemporal comparisons. To control for outlying years, I take the average level of corruption over the period from 1996 to 2010 as the main dependent variable.Footnote ¹²⁰

Results and Analysis

The strong positive relationship between the extent of indirect rule at the time of independence and the perception of corruption in the 1990s and 2000s is illustrated in Figure 1 (r=0.71, p<0.01). Of course, it may be that the high levels of corruption in some states are caused by underdevelopment more broadly and follow a different logic to that outlined above. This cannot be fully ruled out from the data available. However, by controlling for a number of other factors in a multivariate regression, we can see that the relationship between the extent of indirect rule at independence and contemporary corruption still holds.Footnote ¹²¹ It is almost certainly true that economic development and the curtailment of corruption over time went hand in hand.Footnote ¹²² But we would like to know if there is a relationship between the level of development at the time of decolonization (or rather, just before it) and subsequent development. Unfortunately, there is a limited availability of accurate data on gross domestic product (GDP) for many of the smaller colonies. By itself, the log of GDP per capita in 1950 for the twenty-two cases available is negatively associated with corruption decades later (see Table 1, column 2). However, when we include the intensity of direct rule, GDP per capita loses statistical significance and its sign changes, such that those colonies with higher GDP per capita in 1950 have higher levels of corruption in the 1990s and 2000s (column 3).Footnote ¹²³

Figure 1

Direct rule and corruption in the former British Empire Key: BGD Bangladesh; BHS Bahamas; BLZ Belize; BRB Barbados; BRN Brunei; BWA Botswana; CYP Cyprus; FJI Fiji; GMB Gambia; GHA Ghana; GUY Guyana; IND India; JAM Jamaica; KEN Kenya; LKA Sri Lanka; LSO Lesotho; MMR Myanmar; MUS Mauritius; MWI Malawi; MYS Malaysia; NGA Nigeria; PAK Pakistan; SDN Sudan; SGP Singapore; SLB Solomon Islands; SLE Sierra Leone; SWZ Swaziland; TTO Trinidad; TZA Tanzania; UGA Uganda; ZMB Zambia; ZWE Zimbabwe

Table 1

Multivariate Regression Results

Note: The dependent variable is Corruption. Standard errors are reported in parentheses. ***, **, and * denote significance at the 1, 5, and 10 percent levels respectively.

We would also like to know whether some prior factor is causing states to have both higher levels of indirect rule and higher levels of subsequent corruption. Drawing on recent work by Gerring et al., we can include factors that are believed to have caused colonial regimes to have instituted direct or indirect rule.Footnote ¹²⁴ It is likely that indirect rule is partly a function of the size of a given colony. That is, we would naturally expect larger states to have institutionalized indirect rule. This is indeed the case, but indirect rule remains a statistically significant correlate of corruption even when colony size is controlled for (column 8). There is no correlation between indirect rule and factors like ethnic, religious, or linguistic heterogeneity (not shown) or with features of colonization such as pre-colonial population density (column 5) or the size of the settler population (column 6). The most consistent result from Gerring et al.'s study is that more developed pre-colonial states were more likely to have higher levels of indirect rule. A similar conclusion can be implied from Hariri's analysis of pre-colonial state building and postcolonial authoritarianism.Footnote ¹²⁵ I include Gerring et al.'s measure of pre-colonial (or rather, early colonial) stateness. By itself, stateness is positively correlated with corruption in the 1990s and 2000s but its effect loses significance when the more proximate measure of indirect rule is controlled for (column 7). Similarly, there is also a correlation between pre-colonial development (as estimated by Mahoney, Lange, and Von Hau) and subsequent corruption (column 4), but not once indirect rule is controlled for.Footnote ¹²⁶ Essentially, we can infer that more developed pre-colonial states were more likely to have indirect rule, and subsequently became more corrupt (and less developed)as independent states.Footnote ¹²⁷ Thus, we can rule out with some confidence the possible counterargument that more developed colonies simply became more developed, and hence less corrupt, independent states. Finally, we can disaggregate the data into democratic and non-democratic polities. Here I use a level-measure of democracy just prior to the period in which the dependent variable is measured (1995 in this case).Footnote ¹²⁸ The results for democracies and non-democracies are illustrated in columns 9 and 10 respectively. Interestingly, the results are almost indistinguishable across the two regime types. Alternatively, the results also remain robust for the inclusion of a stock variable of democracy, whether binary or continuous, from independence to 1995 (not shown). In short, democratic experience seems to have little impact on the extent of patronage or corruption when we control for the centrifugal legacies of colonial rule.

Although variation in directness of rule is not exogenous to pre-colonial state building, we can have some confidence that the association does not run directly from the characteristics of the pre-colonial states to postcolonial patronage or corruption. It is true that better-institutionalized precolonial states were more likely to experience indirect rule, that such postcolonial states persisted with more decentralized forms of government, and that such states went on to have more contemporary corruption. A reasonable interpretation might be that because empires simply incorporated ‘traditional’ patrimonial state structures through indirect rule, higher postcolonial levels of patronage simply reflect the continuity of traditional political practices in such cases. Where colonial states faced a ‘blank slate’ they could introduce more rational forms of government.Footnote ¹²⁹ There are possible examples such as Mauritius, which was established and populated de novo as a British colony and has since experienced relatively competent government following independence.Footnote ¹³⁰ More systematic historical evidence does not support such a view, however. Within India, for example, there is no evidence of a statistically significant relationship between the geographic extent of direct of rule at the state level and subsequent corruption. India was itself comprised of both directly and indirectly ruled territories.Footnote ¹³¹ In her study of the long-term effects of colonial rule, Iyer finds that at the substate (i.e. district) level in India, direct rule is associated with lower levels of development and public goods provision.Footnote ¹³² This supports the argument that the positive association between direct rule and the quality of governance operates at the national rather than subnational level. That is, indirect rule is correlated with a lower quality of postcolonial governance not because such territories were already ruled by corrupt, patrimonial leaders, but because of the way in which this pattern of indirect governance allowed for the subnational capture of the state during the critical juncture of decolonization. As the section on India above illustrated, it was the leaders of India's directly ruled provinces who pushed the hardest bargain during the institutional and constitutional design of independent India.

These findings are not conclusive; none can directly confirm or disconfirm the theory proposed above. However, the fact that the structural conditions of the late colonial period are systematically correlated with the institutionalization of corruption decades after independence is suggestive. Further empirical tests of the relationship would ideally exploit some exogenous source of variation in centrifugal pressures at the time of state formation, but it is difficult to see how the strict conditions of exogeneity would be met in practice. Nevertheless, when considered in light of the results of other observational and pseudo-experimental research on the relationship between centripetalism and governance, the findings do ring true and have important policy implications.